It's Monday morning, the first week of June, and you're the data analyst at a $40M beauty brand. You already know what your week looks like because the monthly business review is next Thursday, and you’ve repeated the same frantic build up to MBR since you started 2 years ago. By Friday you owe slide 4 with a single number for what Meta paid drove in May, slide 5 a Meta attributed CAC, and slide 6 a recommendation on June budget. However you can’t start any of these slides until you’ve decided which of the four numbers on your monitor is the version of truth the brand is going to operate from.
Meta Ads Manager says paid social drove $1.24M in May. GA4 says $890K. Shopify, filtered by the UTMs the team finally got right last quarter, says $760K. The post purchase attribution survey says around $1.05M. That's a $480K spread on one channel in one month, and the same disagreement will show up again across TikTok, Klaviyo flows, Google Ads, and the new affiliate channel later in the week. By Thursday you'll have reconciled all of it, written the talk track that explains why the May number in this deck isn't the same as the May number in last quarter's deck, and produced the version of the story your CMO will defend in front of the founder. That is a full week, give or take. Once you add in the QBR weeks, a third of your year is going into reconciling the data instead of looking for patterns in it.
Meanwhile, your DMs are stacking up. The brand manager wants to know if the new bundle SKU is cannibalizing the hero. The growth lead wants a read on whether the post Mother's Day dip is seasonality or something to worry about. The CFO wants LTV to CAC by acquisition channel for the board on the 22nd. None of those questions are getting answered this week, because you're in the MBR file.
This cycle of predictable work frustration is the part of the analyst's month most leadership doesn’t take into consideration. Overall, their work breaks into three tiers, but only one of them really needs a human.
Tier 1: data plumbing
The base of the pyramid is the part of the month most growth leads can't see, because none of it ends up in a slide deck. Pulling data from Shopify, Meta, TikTok, Klaviyo, GA4, and Amazon. Normalizing the differences in how each platform reports the same thing. Reconciling discrepancies when the numbers don't agree, which is most of the time.
The size of this tier is well documented even outside DTC. Time spent on data preparation has consistently landed between 45% and 80% of a data professional's working time. Anaconda's State of Data Science survey put data prep at about 45%, and older TDWI executive surveys put it as high as 60 to 80% for more than a third of respondents.
DTC is structurally worse than most industries, because the data is fragmented across more platforms than the average enterprise data team deals with. Most brands find 30 to 60% discrepancies on the same time window when comparing revenue across Meta, Shopify, and GA4. The agency version is worse by an order of magnitude. A 10 person agency managing 25 clients with 3 to 6 active channels each is pulling from 75 to 150 separate platform sources every reporting cycle, and roughly 2 full days of senior analyst time goes into reporting alone before any client gets a recommendation.
With Pulse, this tier is now sitting underneath the data. The pulls, the normalization, and the reconciliation run continuously. Every number traces back to source, time range, and calculation. When platforms disagree, the disagreement is shown for what it is, instead of being papered over by a number that looks more certain than it is. The analyst still owns the answer she gives the CMO. She just stops being the person who has to assemble the raw material every month in order to give one.
Tier 2: repetitive synthesis
The middle tier is the part of the month most legible to a CMO, because it shows up as the Friday report, the weekly performance email, and most of the MBR deck. It is also the work any experienced analyst can do without thinking very hard. Comparing this week to last. Comparing this campaign launch to its baseline. Spotting that the new TikTok creative drove a 22% lift in CPMs and looking around for what else moved at the same time. Rebuilding the same 5 charts the CMO asks for every Friday, then rebuilding them differently when she decides on Wednesday she wants channel CAC trended against contribution margin instead of revenue.
The structure of the weekly questions is the same every week, even though the questions themselves aren't easy. Most weeks the answer is "nothing changed enough to act on." A smaller number, "something moved, here's the most plausible cause." A very small number, "something changed materially, consider acting before the next cycle." A senior analyst with 2 years of context inside a business recognizes which of those weeks she's in inside the first hour. The other 6 hours that go into the Friday report are formatting and cross checking.
This is also where the rabbit holes start. The analyst spots an unexpected dip in Klaviyo flow performance and spends 11 hours pulling cohort data to figure out whether it's deliverability, segment drift, or a real engagement problem. Sometimes that's the most valuable work she does all month. Often the dip turns out to be an artifact of a metric definition that changed three weeks ago and was never real to begin with. The rabbit hole and the genuine investigation look identical in the first hour.
With Pulse, this tier is also sitting underneath the data now. The comparisons run continuously, the flagging of "something moved" runs continuously, and the proposed causes come with the supporting math attached, so the difference between a real signal and a metric definition artifact shows up at the start of the investigation rather than at the bottom of it. What the analyst does on Friday is read what surfaced, judge what's worth elevating, and write the version of the story her CMO will read on a phone between meetings.
Tier 3: real judgment
The top of the pyramid is the part of the month most invisible to growth leads, and the only tier that stays with the human. Deciding what the moved numbers mean for next quarter's plan. Choosing which of the 4 interesting things this month are worth investigating now and which can wait. Deciding when the data isn't enough to answer a question and spending the budget anyway because waiting is more expensive than being sure. Telling the CMO her favorite channel is no longer pulling its weight, and being right enough that she trusts the read the next time too.
This is also where the deeper work lives, the work the analyst was actually hired for. The double click into a customer cohort that's started behaving differently and is going to reshape the retention math by Q4. The trend that's too quiet to see in a weekly chart but obvious if you pull 18 months of data and segment it correctly. The half day spent figuring out whether the brand should keep paying its agency for the same media plan they've been running since 2024. None of that happens during MBR week, because MBR week is plumbing and synthesis. Most of it doesn't happen during the other three weeks either, because the ad hoc requests stack up and the deep work gets pushed to next month.
Move tier 1 and tier 2 underneath the data and the analyst's month opens up. The MBR deck is essentially drafted by Wednesday of the first week. The ad hoc questions get answered the same day they're asked. And the trend hunting, the cohort work, the agency value review, the analysis that compounds over time, finally has room to happen.
We're deliberate about where the system stops. The product is read-only and advisory by design. Pulse surfaces what's happening in the data, traces every claim back to source, points the analyst at where to look next, and makes recommendations with confidence levels attached. It does not pause a campaign, move a budget, or take action inside any connected platform. The recommendation is the output, and the decision belongs to the room. Most of the analytics tooling coming to market right now is leaning the other direction, into autonomy and "we'll just do it for you." That's the wrong framing for marketing decisions at the growth stage, where the decisions that matter are full of context that isn't in the data: the competitive set, the seasonality of the category,and the founder's tolerance for risk this quarter. A system that acts on the analytics in isolation takes action in opposition to the strategy.
What this changes about how to staff growth
A US based mid level retail marketing data analyst averages roughly $86K to $129K in base salary in 2026, with seniors running higher. Loaded for benefits, payroll taxes, equipment, software, and overhead running that cost 25-40% higher. For a DTC brand at $5M to $100M in revenue, that's a real number, and it's the number every growth lead is weighing against the alternative of operating without analytical horsepower in the room.
The question this year isn't whether to hire another analyst. It's whether the one you have is spending her month in tier 3, where a human is genuinely irreplaceable, or whether two thirds of her month is going into tier 1 and tier 2 work that doesn't need her judgment. If it's the latter, the answer isn't to hire a second analyst to do more of the same. It's to move tier 1 and tier 2 underneath the data and give the analyst back the part of her month she was actually hired to do.
FAQ
How much of a DTC data analyst's time goes to data preparation?
Industry surveys consistently put data preparation at 45 to 80 percent of a data professional's working time. DTC is structurally worse than most industries because the data is fragmented across more platforms — most brands see 30 to 60 percent revenue discrepancies on the same time window when comparing Meta, Shopify and GA4.
What are the three tiers of a DTC analyst's monthly work?
- Tier 1 — data plumbing. Pulls, normalization, and reconciliation across Shopify, Meta, Klaviyo, GA4, TikTok and more.
- Tier 2 — repetitive synthesis. Week-over-week comparisons, the Friday report, the MBR deck, and most ad-hoc questions.
- Tier 3 — real judgment. Deciding what the moved numbers mean for next quarter, double-clicking into cohorts, and recommending action under uncertainty. Only Tier 3 truly needs a human.
What does Hologrow's Pulse replace, and what stays with the human?
Pulse moves Tier 1 (data plumbing) and Tier 2 (repetitive synthesis) underneath the data: the pulls, normalization, weekly comparisons and "something moved" flagging run continuously, with lineage and confidence on every number.
Tier 3 — judgment, prioritization, and deciding when the data isn't enough — stays with the analyst. The product is read-only and advisory by design: it does not pause campaigns or move budgets.
Why doesn't Hologrow take action automatically inside ad platforms?
Marketing decisions at the growth stage are full of context that isn't in the data: the competitive set, category seasonality, the founder's risk tolerance this quarter. A system that acts on the analytics in isolation can take action in opposition to the strategy.
Pulse surfaces what's happening, traces every claim back to source, points the analyst at where to look next, and makes recommendations with confidence levels — but the decision belongs to the room.
